TY - GEN
T1 - PixelSynth
T2 - 18th IEEE/CVF International Conference on Computer Vision, ICCV 2021
AU - Rockwell, Chris
AU - Fouhey, David F.
AU - Johnson, Justin
N1 - Funding Information:
We thank Angel Chang, Richard Tucker, and Noah Snavely for allowing us to share frames from their datasets, and Olivia Wiles and Ajay Jain for easily extended code. Thanks Shengyi Qian, Linyi Jin, Karan Desai and Nilesh Kulkarni for the helpful discussions.
Publisher Copyright:
© 2021 IEEE
PY - 2021
Y1 - 2021
N2 - Recent advancements in differentiable rendering and 3D reasoning have driven exciting results in novel view synthesis from a single image. Despite realistic results, methods are limited to relatively small view change. In order to synthesize immersive scenes, models must also be able to extrapolate. We present an approach that fuses 3D reasoning with autoregressive modeling to outpaint large view changes in a 3D-consistent manner, enabling scene synthesis. We demonstrate considerable improvement in single-image large-angle view synthesis results compared to a variety of methods and possible variants across simulated and real datasets. In addition, we show increased 3D consistency compared to alternative accumulation methods.
AB - Recent advancements in differentiable rendering and 3D reasoning have driven exciting results in novel view synthesis from a single image. Despite realistic results, methods are limited to relatively small view change. In order to synthesize immersive scenes, models must also be able to extrapolate. We present an approach that fuses 3D reasoning with autoregressive modeling to outpaint large view changes in a 3D-consistent manner, enabling scene synthesis. We demonstrate considerable improvement in single-image large-angle view synthesis results compared to a variety of methods and possible variants across simulated and real datasets. In addition, we show increased 3D consistency compared to alternative accumulation methods.
UR - http://www.scopus.com/inward/record.url?scp=85120601645&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85120601645&partnerID=8YFLogxK
U2 - 10.1109/ICCV48922.2021.01384
DO - 10.1109/ICCV48922.2021.01384
M3 - Conference contribution
AN - SCOPUS:85120601645
T3 - Proceedings of the IEEE International Conference on Computer Vision
SP - 14084
EP - 14093
BT - Proceedings - 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 11 October 2021 through 17 October 2021
ER -